name: "gptj"
backend: "python"
max_batch_size: 128

input [
  {
    name: "INPUT__0"
    data_type: TYPE_INT64
    dims: [128]
  }
]

output [
  {
    name: "OUTPUT__0"
    data_type: TYPE_INT64
    dims: [2048]
  }
]

instance_group [
    {
        kind: KIND_MODEL
        count: 1
    }
]
dynamic_batching {
  preferred_batch_size: 128
}
parameters: {key: "COMPILED_MODEL", value: {string_value: "EleutherAI-gpt-j-6B-bf16-local"}} 
parameters: {key: "NEURONX_DUMP_TO", value: {string_value: "gpt-j-6b-artifacts-tp4"}} 
parameters: {key: "NEURON_CORE_START_INDEX", value: {string_value: "0"}} 
parameters: {key: "NEURON_CORE_END_INDEX", value: {string_value: "11"}} 
parameters: {key: "NUM_THREADS_PER_CORE", value: {string_value: "1"}}